Hello Robert,
Thursday, May 23, 2002, 11:47:26 AM, you wrote:
>> Well, this will fix the problem, but String::substr() will
>> still cause a crash if for some reason a NULL pointer is
>> passed to it! Btw have you read my mail till the end ?
RC> Yes. The String++ implementation currently uses the <string> std class
RC> to implement substr. From a quick inspection the fault is in the
RC> string() constructor not returning an empty string when NULL is passed
RC> to it.
No! The string constructor which uses const char * as its argument - there is
no check if the pointer being passed is NULL and later strlen () is
called on the pointer to determined the size of the C string being
passed... Here is the exact flow:
1. The constructor
basic_string (const charT* s)
: dat (nilRep.grab ()) { assign (s); }
2. The assign call
basic_string& assign (const charT* s)
{ return assign (s, traits::length (s)); }
3. The call to traits::length()
static size_t length (const char_type* s) { return strlen (s); }
The crash happens in 3 - a string object is never constructed at all
and string::substr() is never called. This is a bug of the string
implementation in 2.95.3. I'am aware of this because I've done alot of
usage of strings lately and was bit by this. I avoid this by calling
string::assign () instead of the string (const char *) constructor
which is safe.
--
Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple
Bug reporting: http://cygwin.com/bugs.html
Documentation: http://cygwin.com/docs.html
FAQ: http://cygwin.com/faq/
